Agent Harness Engineer

JAPAN AI (a group company of Geniee, Inc., TSE Growth Market) is hiring an Agent Harness Engineer to design and implement the Agent Harness — the execution engine, orchestration, guardrails, memory, and model routing — that enables AI agents to operate safely, quickly, and reliably. This is the infrastructure that transforms an agent from "works in a demo" to "trusted in production."

WHAT YOU'LL DO

• Deeply understand LLM/AI agent operating principles and design/implement the execution engine (Graph Runtime / State Machine).
• Develop an Agent SDK used by ~120 in-house engineers.
• Build the guardrail / policy execution engine that safely controls agent behavior and LLM outputs.
• Handle AI-specific system design such as model routing across multiple LLM providers, context management, and memory infrastructure (long-term memory, working memory, sessions, checkpoints).
• Integrate RAG and optimize inference pipelines and orchestration systems for cost and latency.
• Maintain ≥99.9% platform uptime.

TECH STACK
Python, Go, TypeScript/React, Next.js, GCP, Docker, Kubernetes, Terraform, Kafka, Pub/Sub, Prometheus, Grafana, OpenTelemetry. Reference frameworks: LangChain, LangGraph, AutoGen. AI dev tools provided: Claude (Claude Code MAX Plan), Cursor, ChatGPT, Devin.

REQUIREMENTS

• 5+ years backend engineering experience.
• Production Python development.
• Distributed systems design and implementation.
• LLM / AI agent production system experience.
• Business-level Japanese (required); business-level English.

NOTES
Japan residency required. Housing allowance up to ¥30,000/month; development tool support; reviews and bonuses twice yearly; stock options.

[partial — verification agent will extract full description]